To kill the Communication Center (the very first thing I do after a revit install) in Regedit replace all the (1) with (0) see image.

Then rename WSCommCntr2.exe to AAWSCommCntr2.exe
( C:\Program Files\Common Files\Autodesk Shared\WSCommCntr\lib\AAWSCommCntr2.exe)


Then it is put to sleep.

When normally start revit and close it the popup doesn't happen.
I have removed all our add-ins and preformed a clean install (I have taken a shortcut with the clean install script not manually)
Then I have killed Communication Center.

The popup still remains (like vermin)

Edit:
The Windows Logs are now clear from errors.
I have also added my journal file I'm not too good at reading them.
I have tested it with my computer it gives me the error also. I will test this journal file also on a other computer.

Well I have been a bit busy so no update for a while.
I have also tested it on RST and RME. RST can't run the journal to completion and then gets the same popup and rme works normally but at the end still gets the popup.

Sander,
the problem with RST is known, but hopefully with some much appreciated help from Scott B at the Factory I will finally be able to put that one to bed soon.
The other issue I am not sure about. I have not had any problems like this in testing, and I actually don't even have CommCenter disabled on my VMs. Hopefully someone can figure out what is going on from your posted journal. I will take a look tonight.

Gordon

All,
I have been wracking my brains trying to get AUBench to work in RST, to no avail. Then I had an epiphany. Rather than a single test that creates and exports, perhaps two tests. The Model Creation journal is just the AUJournal up to the point of Rendering, which is the point of failure in RST. Then a second journal that opens a prepared file and does the rendering and export tasks. We will end up with a single test (via multiple journals that are run automatically via VBScript) that can run on any vertical and allows direct comparisons. I am proposing that this new benchmark tool be called RFOBenchmark, and Phil Read has given his blessing for using the original AUBench as the seed. Now the new approach will mean that direct comparisons between RFOBench and AUBench will be less than apples to apples, but with caveats these comparisons will still be useful. However, in breaking with AUBench, some opportunities present themselves.

1: Adjustments to the AUBench portions. Things like a higher quality rendering, or exporting to PNG rather than TIF (I feel like PNG is the more commonly used format now, so PNG would be more "real world").
2: Additions to the AUBench portions. For example adding a DWG export section. Or DWF. Or perhaps adding other tests in the Model Creation portion? Is there anything
3: The addition of a hardware Acceleration benchmark. I have an approach that I am confident actually tests hardware accelerated performance, which I think is a long missing piece of data when making hardware decisions.

So the question is, do we want to keep all the AUBench portions the same, so that we can compare to historical data? Or do we want to make the new RFOBenchmark more comprehensive, even if it does make comparisons with old numbers impossible? And if we want to make changes, what are they? As I mentioned, I lean towards a higher quality rendering, PNG as export format, and DWG export. The AUBench addresses groups, and updating a family, as well as creating views. Linking DWGs and/or RVTs might be useful.
I would also love to add a Print section, but I don't see how considering everyone has different printers and PDF print capabilities. If nothing else I think the new approach is modular enough that making an in-house Print test would be easy enough. I may also try to test various PDF solutions.
So, if anyone has any suggestions here I would love to hear them. In the meantime, I will be working on finalizing a few things and hopefully we can roll out an all new RFOBenchmark next week!

Thanks everyone.

Gordon

Hi Gordon
I am all for the new RFOBenchmark. As for historical data comparison, not bothered. I spec'd my recent custom I7 X980 24gb pc for rendering and felt that the AUBench didn't really push the render side far enough to get real results.
From what I saw of the AUBench results there is quite a broad range of machines between the RFO users, and I am sure more members will jump on board for some interesting comparisons.
I am all for pushing the thing to the max. and getting some real results.

All, the new RFO benchmark is coming along nicely, with only a minor tantrum thrown by RST. Of course. Hope to get that addressed tonight.

So I have added a few things, including a full DWG export, a higher quality render, and a hardware acceleration benchmark. Net result is the full test now runs 10-20 minutes! I am not overly concerned there but I want to get the group's thoughts on such a long benchmark.

Some other improvements include
Run from any location. The benchmark will work from the C drive, or a network drive, anywhere.
No need to turn on Press & Drag. And it will be off again at the completion of the benchmark.
Single test, for all three versions of Revit.

I will be finalizing everything in the next few days, then making a download available to a few people to test. My hope is to not have the constant updates, but rather have a final working version as the first and only post. Look for something by the end of the week.

Gordon

P.s. Results look like this...


Model creation and view export benchmark
6.00 opening and loading the custom template
11.00 creating the floors levels and grids
31.00 creating a group of walls and doors
46.00 modifying the group by adding a curtain wall
25.00 creating the exterior curtain wall
12.00 creating the sections
13.00 changing the curtain wall panel type
4.00 save the project
2.00 open the project
46.00 export all views as PNGs
172.00 export all views as DWG
368.00 Total

Render benchmark
267.00 render

GPU benchmark
65.00 with hardware acceleration
245.00 without hardware acceleration

That sounds awesome Gordon - Am especially thrilled by the more thorough rendering
One tiny thought though... the "Export to DWG" seems to account for half the GPU benchmark - Maybe it would be better to make the model more advanced, and only export 20% of the views to DWG ? Just a thought... and keep up the great work :whiskey:

PS: Let me know if you need a beta tester

Well, I started with fewer, but then I realized that I now had two "Export" tests, PNGs and DWGs, and I thought I would want to be able to compare relative speed of each. If I had a different export set for DWGs, would that make it impossible to really compare? But at the price of a much longer test, is it worth it? I remain on the fence.
If nothing else, the ridiculously slow DWG export tells me that Autodesk REALLY needs to put some effort into this. It is a single thread doing the work, and yet it is a textbook "Embarrassingly Parallel" problem. Being able to say "Hey, Factory, PNGs are four times as fast to export as DWGs, what gives?" might help get some development effort focused here. Indeed, I am going to also add (for my own use) a DWF export as well. I am hoping to make the whole thing very modular, so someone can add whatever specific tests they want. For example, I might test in house doing some standardized Sync with Central task to see just how bad our network is right now. Or to compare file based SwC vs RevitServer based SwC. Or perhaps compare motherboard based gigabit network connections with dedicated NICs. Copy monitor seems like an option too.

However, adding to the model is probably the hardest part of the journal. I have managed to get as far as I have because I had Phil's original journals to work with. Everything I have added is flavors of exports, plus addressing the relative paths to things so the folder can move around. I even had to punt to a stock copy of the file to get this far. Nothing I did would make the old journal work in all three flavors.
Of course that moving folder thing worked on my machine at home just fine, but on a real (Windows) network it is now failing. Gah!

We shall see how it goes from here. Hopefully I will need some beta testers soon!

Thanks,
Gordon

Our clients ofter work with revit files over 100 mb I would also like to have a benchmark that can mimic that I don't know how other people feel about this?
I don't know if it possible to start with a larger model and edit this model change edit wall types and such I don't know if that is easier?